First working version

app.py

@@ -34,8 +34,9 @@ title = """
 def get_bg_image(prompt):
     images = pipe(prompt)
     print("Image generated!")
-    image_output = images.images[0] if not images.nsfw_content_detected[0] else Image.open("nsfw_placeholder.jpg")
-    return image_output
+    image_output = images.images[0]
+    image_output.save("img.png")
+    return "img.png"
 
 def get_music(prompt):
     spec = pipe2(prompt, height=512, width=512).images[0]
@@ -49,7 +50,9 @@ def infer(prompt):
     image = get_bg_image(prompt)
     audio = get_music(prompt)
     return (
-        gr.make_waveform("output.wav", bg_image=image, bars_color=random.choice(COLORS)),
+        gr.make_waveform(audio,
+                         bg_image=image,
+                         bars_color=random.choice(COLORS))
     )
 
 css = """
@@ -84,7 +87,7 @@ with gr.Blocks(css=css) as demo:
                                 elem_id="prompt-in",
                                 label="Enter your music prompt")
       with gr.Row(elem_id="btn-container"):
-        send_btn = gr.Button(value="Send", elem_id="submit-btn")      
+        send_btn = gr.Button(value="Send", elem_id="submit-btn")   
       send_btn.click(infer,
                      inputs=[prompt_input],
                      outputs=[gr.Video()])